Pandora's Magic Box
Restoration of the music's dynamic range in historical recordings
by: Pierre A. Paquin
We are now able to enjoy great recording of the distant past, thanks to the efforts of several successful and dedicated engineers who have had at their disposal the original master discs and tapes of those recordings they wish to restore to present-day listenablity. Noteworthy among the major recording companies are BMG with the cast RCA Victor catalog. EMI by way of Testament and quit a few other engineers with variable success using sources other than the original. However, there is one major annoyance caused by some of these restorations namely an unwillingness to infuse some of the dynamic range, which was removed in many of these old recordings by the original engineers during the actual recording sessions. One would expect a dynamic range process to be much more possible these days with all sorts of electronics available within the analog and digital domains.
While listening to remarkable restorations of few the few records of Irish soprano Margaret Sheridan made at La Scala in the late 1920s I became convinced that the recording process for one major company began to degrade as time went on, especially beginning in 1943 and ending in 1951. These La Scala recordings as restored successfully by both Romophone (F 89001-2) and Time Machine (F 0100) expose how great an electrical recording could be 70 years ago. These Italian recordings were made by HMC featuring the great soprano in a complete Madame Butterfly as well as duets with Toscanini's favorite tenor, Aureliano Pertile. The La Scala Orchestra was conducted by HMV's Italian music director, Carlo Sabajno. String portmento was a way of life in the La Scala Orchestra in the 1920s, as these recordings show, and the sonics are incredible for the time.
These old records seem to verify the idea that whoever was at the recording 'console' way back then had an idea about the music being etched by needles on to wax master discs. Occasionally there is come evidence of overload - hence the necessity to issue some of these La Scala sides as dubbings - but overall the dynamic range (soft to load and load to soft) is intact and very much up to date. The original engineers would use a manual technique called 'gainriding' to make sure that the pianissimos were not buried in surface noise and most of the fortissimos were brought down in volume intensity so as to not exceed the 78rpm recording limitations. This practice, which was very necessary during those old recording sessions would be successful musically only if the engineer knew the music by way of listening to other recordings or being able to read the musical score well in advance of the sessions.
Next on the listening agenda were the 1928 Beethoven Pastoral Symphony on the original 78rpm discs (Victor set M50) and the 1930 Tchaikovsky Pathtique (RCA 0902660920-2) both with the Boston Symphony Orchestra under Serge Koussevitzky. These recordings were engineered with remarkable finesse, at least in terms of dynamic range. With the Pathetique CD is that remarkable performance of Tchaikovsky Romeo and Juliet etched in 1936, also with the BSO under Koussevitzky, with a little less success in overall sound even though six years had passed since the Pathetique had been recorded. Next, on with the BSO/Koussevitzky Tchaikovsky Fourth Symphony recorded a few years later not bad at all were dynamic range is concerned but not incredible in terms of performance. (Would that the BSO could play like that today.) Then in 1943, something happened to RCA Victor engineering. It seems that the engineering staff no longer needed to have any idea about the music to be recorded, as a 'magic box' was introduced into the recording chain of electronics. It appears that anyone could be a recording engineer of great music with that box in place. Dynamic range of the music now took a back seat as this primitive compressor automatically replaced the need for any judicious and necessary manual 'gainriding' of the console master volume control. One of the first recordings to fall victim was the great BSO/Koussevitzky Appalachian Spring Suite (Victor 11-9129/31) then on with their legendary 1946 performance of Schubert's 'Unfinished' (Victor 11-9082/4) up to the last Boston Symphony recording of Koussevitzky in 1950: Sibelius's Second Symphony (Victor LM1172). The compression was so bad in this recording that it was overcome musically only by the performance itself. The 1951 Schumann Spring Symphony with Munch, now in vinyl heaven (Victor LM1190) also comes to mind, as well as the same conductor's great performance of Schumann's Genoveva Overture with BSO (RCA Victor 0926 60082-21). Other examples of 'automation' are all the Toscanini recordings done in studio sessions during the same period. Think of the sessions for the Rossini Overtures from the mid-1940s (the NBC broadcasts were much better engineered, as the recording staff were quite different).
The recording session that forced RCA Victor engineers to get back to learning the score was undoubtedly that for the Toscanini Verdi Requiem (RCA 74321 72373-21). Ironically, it was the Requiem not only for Manzoni but also for the 'magic box' compressor. It was a broadcast performance but RCA Victor engineers were given the responsibility of doing the recording instead of the usual NBC broadcast crew. So ill prepared were the Victor engineers in coping with the dynamics of Verdi's 'best opera', with Toscanini on the podium shaking his fist for more volume from the thunderous if not infamous massive square bass drum designed especially for this Carnegie Hall performance that the final product was a patchwork of rehearsal and performance takes. Plainly evident to the ears is the magic box trying desperately to do its automated duty. Yet this performance is so dramatic in every way that it overcomes the very poor if not stupid technical work done by the Victor engineers in 1951.
During this seven -year period those incredible performances with Monteux and the San Francisco Symphony Orchestra fell victim to this careless (I could care less!) recording technique. All can be heard within the Monteux CD edition on RCA. (I shall not comment on the 'front end' distortion: evident on the Monteux/BSO Tchaikovsky Fourth Symphony of 1959. That is another story.) Remastering here creates a rather noise-free listening experience throughout the CD set but what could have been done to fix the effects of that wretched compressor in order to retrieve the dynamic range of the performances.
It seems quite logical that if we want to restore and preserve these RCA Victor performances, the huge scrapheap of discarded electronics should be thoroughly searched in order to find the 'magic box' compressor. Its horrid thumping and pumping compression parameters should be analyzed and computer sound stations should have a special programme developed in order to counteract these ill effects in engineering a truly complete restoration of these fine performances.
The RCA Victor recording which must fixed is the Sibelius Second Symphony with Koussevitzky, as it can best demonstrate the wholesale use of the 'magic box'. Moving on to the 1947 Munch/BSO Beethoven Seventh Symphony (Victor LMI ?34), what a performance that is and it is one of the very first RCA recordings made with analog tape: the first RCA recording of the Boston Symphony Orchestra with 'Le Beau Charles' as he was called by the French-speaking ladies in the Symphony Hall audience.
Pierre Paquin has been a recording engineer for more than 30 years. His best work is with symphony orchestras and choruses. He now helps his wife operate Sound Dynamics Associates, a recording company based at Cape Cod Massachusetts. See his website at http://www.sd-associates.com
"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche